
Tech • AI • Robotics • Game
Alphabet shares fell sharply after an earnings miss exposed growing pressure on Google Search monetization even as Google Cloud and Gemini expanded rapidly, while a separate report intensified scrutiny of efforts to route Nvidia chips toward China.
Alphabet shares dropped about 12.5% after reporting revenue of $127.6 billion, slightly below the $128 billion consensus. Earnings per share came in at $2.12 versus expectations of about $3.01, a steep miss that revived investor concerns about the company’s core advertising engine.
Part of the earnings weakness came from an $8.7 billion loss on equity investments linked to a decline in SpaceX valuation, which cut earnings by about 55 cents per share. Even excluding that hit, adjusted EPS would have been roughly $2.67, still below Wall Street forecasts and still weak enough to unsettle investors.
Google Cloud posted revenue of $30.7 billion, roughly double prior levels, while operating profit rose to $11.1 billion. Operating margin reached 36.2%, underlining how far the business has come since earlier internal debates over whether it should even remain part of the company.
The cloud backlog reached $561 billion, with about 55% expected to be recognized as revenue within two years. That suggests unusually strong near-term visibility for a business benefiting from demand for AI infrastructure, including cloud services and custom TPUs.
Search advertising growth cooled to 12%, down from 17% in the previous quarter. The company is seeing rising AI query volume, but AI Mode appears to generate less revenue than traditional search because conversational responses offer fewer obvious places to load a page with ads.
Traditional search has long been optimized to monetize nearly every part of the results page, from sponsored links to shopping units. AI-generated answers compress that experience into a cleaner response, creating a strategic challenge: how to insert profitable advertising without degrading the product.
Google signaled about $195 billion to $205 billion in AI infrastructure spending for 2026, with expectations for a further increase in 2027. Third-quarter spending alone reached $51.4 billion, reflecting the scale of the race to build data centers, train models and secure enough power and hardware.
Gemini surpassed 1 billion monthly active users, and its API is processing 32 billion tokens per minute, up 40% quarter on quarter. The launch of Gemini 4 Carbon reinforced the company’s argument that owning frontier models matters not only for external sales but also for internal products such as Search and YouTube.
The spending race is not limited by demand alone. Microsoft reported capital expenditures of $38.37 billion, below estimates of $45.9 billion, while Google’s own capex also came in light versus expectations, pointing to delays tied to construction, permitting, power and deployment rather than any easing in appetite.
A report said U.S. officials concluded that Alice Wong, an executive at a Singapore-based data center company, was passing intelligence to Beijing and had developed a close relationship with Nvidia chief Jensen Huang. The allegation heightened concern that intermediaries in Malaysia or Singapore could be used to gain access to advanced AI chips restricted from reaching China.
The enormous value of advanced Nvidia GPUs has created strong incentives for smuggling, relabeling and transshipment through third countries. That has made countries with rapidly expanding compute capacity, especially where local demand appears out of proportion to the broader economy, a focus of scrutiny for U.S. policymakers.
The latest results showed Alphabet balancing exceptional AI and cloud momentum against a more fragile transition in search advertising. At the same time, the contest for AI leadership is spilling into geopolitics, where access to advanced chips is becoming as strategic as the models they power.
Ask a question